Real-Time peaks/dropouts with VEP instrument tracks connected to a slave

I and other folks over at the vi-control forum are having some strange random real-time peaks and dropouts by record enabling and playback Vienna Ensemble Pro 6 instrument tracks connected to a slave. The peak meter ranges are going from around 10% to 100%, very unstable, jerking up and down during playback of a very light project, never mind if those tracks have anything to play or not, it peaks as long as they are record enabled.

Loading the same library in VEP locally on the master PC doesn’t introduce any problems with C10, only via Ethernet, streaming from a slave. Tried with two different slaves and have the same issue with both.

I lost days troubleshooting to finally go back to 9.5 as there are NO SUCH PROBLEMS with the previous version.

LatencyMon doesn’t show the problem. I tried many things, AG on/off, fresh win install, different video drivers (AMD), pref rebuild (this introduced even more peaks), couldn’t find anything locally related to the problem.

Hoping that this will be fixed as I love some of the new features and would like to use them.

shomynik / others - do you know if this affects both VST2 and VST3 versions? Would you have time to do a quick test if not sure?

I always use vst3 as per multiple recommendations. But a good call. Will try.

Same here wtih Cubase 10 and VEPRO on slave PC.
Big CPU spikes upon playing a project
Crashing when hitting the record button
Back to 9.5 for now

Just checked the vst2 version of the VEP plugin and yes, same issue there.

I loaded only one VEP vst2 plugin via instrument track, and with VEP buffer set to 2, real-time peak jumps are small enough to don’t cause the dropouts, but they are there. But as soon as I change the buffer to 1, it starts to jump full scale, from very minimum (as I have only one track loaded in the project) to a full tilt, totally random.

However, there is no such behavior on 9.5 with either buffer setting.

Thanks for checking, shomynik. Weird that this only affects slaves. Hopefully at least that narrows down considerably what it could be from a code perspective.

Might be also worth posting these findings over on the VSL board.

Another great call Guy, thanks. You’re right. Will do.